View Full Version : Meticulous detail
I have just spent 4 days in my house judging the sonic differences between I/U transistors in the SW1X DAC. Slawa supplied the DAC and 6 sets of PNP resistors with a circuit board set up so that each could be plug and played.
Well those sceptics that claim all components sound the same can turn away now.
My criteria was not about Hi Fi qualities such as extended bass, or fuller mid range, but more about was I really enjoying the sound, would it keep me listening late into the night, and if there were 6 DACs separately fitted with just one of the pairs of trannys, which would I buy?
Slawa had pre selected the six trannys from a host of candidates ranging from rare 1960s stuff to modern equivalents, some silicon and some germanium. I wont go into detail, but suffice to say that marking out of 10, the worst got 4 and the best got 9. Differences were not subtle.
I use this just as an example of how one component in a circuit can influence the sound of the final product. Slawa tests most components in this way
sq225917
07-09-2017, 20:46
I'm more intrigued how the circuit was compensated for the electrical differences between the transistors? Was it just a straight transistor swap or were they mini-circuits?
How did Germanium score?
As usual, they beat England on penalties! :lol:
Sorry, as you were.
Marco.
How did Germanium score?
The preselection process had only brought 1 germanium to my testing, but it did finish in second place.
I'm more intrigued how the circuit was compensated for the electrical differences between the transistors? Was it just a straight transistor swap or were they mini-circuits?
I wish I had the skills to answer that, but I just listen to music. I'm sure Slawa may be along to help explain the technical stuff.
As usual, they beat England on penalties! :lol:
Sorry, as you were.
Marco.
Love it.
I'm more intrigued how the circuit was compensated for the electrical differences between the transistors? Was it just a straight transistor swap or were they mini-circuits?
I spoke with Slawa briefly about this yesterday. Apparently its a bit like valve rolling, except in this case its with I/U transistors. This is achievable if the surrounding 'circuitry' is vastly over specified as it is in his equipment, and the trannys are selected to do a similar job in that circuit.
Arkless Electronics
09-09-2017, 14:37
:rfl:
Haselsh1
09-09-2017, 14:53
How did Germanium score?
Considerably better than Francium ;)
Haselsh1
09-09-2017, 14:53
As usual, they beat England on penalties! :lol:
Sorry, as you were.
Marco.
:eyebrows:
I spoke with Slawa briefly about this yesterday. Apparently its a bit like valve rolling, except in this case its with I/U transistors. This is achievable if the surrounding 'circuitry' is vastly over specified as it is in his equipment, and the trannys are selected to do a similar job in that circuit.
Greetings,
The pair of transistors for the I/U conversion stage are actually bipolar NPN type. The NPN type of transistor has a similar structure to a triode valve (anode=collector, grid=base and cathode=emitter) except having no filament and using a solid state semiconductor material such as Germanium or Silicon instead of vacuum. Their bias is only a couple of mA, while the current ability of the transistors and the discrete shunt power supply easily approaches or exceeds 1A.
Since all the NPN bipolar transistors have 3 pins one can easily replace them in a matter of seconds by installing 3 sockets on the PCB of the DAC. That provides the possibility to voice any NPN transistor in a given circuit and to customise the sound to its own liking.
S
Lawrence001
15-09-2017, 13:45
Sounds simple, why do I have the feeling someone will respond to say it's not quite that easy...
Sent from my NEM-L51 using Tapatalk
Sounds simple, why do I have the feeling someone will respond to say it's not quite that easy...
Sent from my NEM-L51 using Tapatalk
Rolling transistors is simple. With a little bit of practice is almost as simple as rolling tubes.
S
Arkless Electronics
15-09-2017, 15:45
Sounds simple, why do I have the feeling someone will respond to say it's not quite that easy...
Sent from my NEM-L51 using Tapatalk
Let's just say my earlier ROFL was NOT due to Marco's "joke"....
....and on that bombshell, goodnight.
RothwellAudio
15-09-2017, 16:15
While it's true that many transistors have enough in common to be interchangeable and in many circuits feedback will iron out any differences, this does smack of a "suck it and see" approach or a hit-and-miss or trial-and-error approach.
Besides, transistors are entirely the wrong shape to be good for rolling - at least valves are cylindrical :lol:
While it's true that many transistors have enough in common to be interchangeable and in many circuits feedback will iron out any differences, this does smack of a "suck it and see" approach or a hit-and-miss or trial-and-error approach.
Besides, transistors are entirely the wrong shape to be good for rolling - at least valves are cylindrical :lol:
It may be true in many circuits, however we do not employ any kind of feedback loop(s) in our circuits. It also applies to the pair of NPN transistors in the I/U conversion stage.
I must admit that it is a trial and error approach, though. That is also a part of R & D, which is costly & time consuming but can sometimes be very rewarding. I would recommend to try it sometimes.
Arkless Electronics
15-09-2017, 16:37
:popcorn:
Lawrence001
16-09-2017, 19:13
I know op amp rolling is very popular in headphone amps (although that doesn't mean it's a good idea). Is there any reason transistors wouldn't work the same? I think this is a technical discussion worth pursuing.
Sent from my NEM-L51 using Tapatalk
I know op amp rolling is very popular in headphone amps (although that doesn't mean it's a good idea). Is there any reason transistors wouldn't work the same? I think this is a technical discussion worth pursuing.
Sent from my NEM-L51 using Tapatalk
Absolutely, the only but very significant difference is that we do not use op-amps as those integrated circuits are very complex in nature, containing dozens of transistors, resistors, diodes and negative feedback loops- all of which is responsible for degradation of sound beyond belief. There is a huge difference (like night and day) between a couple of op-amps containing around 20-40 transistors each and a pair of specially selected, quality transistors for the I/U followed by a powerful, low internal resistance triode running in single ended class A without any negative feedback loop.
http://sw1xad.co.uk/technology_post/frequently-asked-question-faq/
One could also replace the I/U transistor with a carefully selected resistor and live with just one triode instead but then one would loose out on additional resolution and dynamics.
If that is desired then we offer that kind of solution too.
S
Arkless Electronics
17-09-2017, 12:26
I know op amp rolling is very popular in headphone amps (although that doesn't mean it's a good idea). Is there any reason transistors wouldn't work the same? I think this is a technical discussion worth pursuing.
Sent from my NEM-L51 using Tapatalk
No you can't do it with transistors in the same way as op amps.
No you can't do it with transistors in the same way as op amps.
Why not?
Check this http://sw1xad.co.uk/transistor-rolling-for-the-iu-conversion-option/
Rolling transistors is almost as easy as rolling valves or op-amps or any components that have pins and could be installed on sockets.
As with any component that has pins it is paramount to make sure that it is pin compatible and to observe the correct orientation.
S
Arkless Electronics
17-09-2017, 17:01
Why not?
http://sw1xad.co.uk/transistor-rolling-for-the-iu-conversion-option/
Rolling transistors is almost as easy as rolling valves or op-amps or any components that have pins and could be installed on sockets.
As with any component that has pins it is paramount to make sure it is pin compatible and to observe the correct orientation.
S
Because they come in NPN and PNP, germanium has a Vbe of 0.2V but silicon 0.7 (roughly), they have different voltage ratings, max current, power ratings, Hfe, Ft, internal capacitances etc. A silicon transistor will not work in a circuit designed for germanium and vice versa (other than certain types of emitter follower). The different Ft and capacitances may cause oscillation, either locally, or in the case of a feedback amplifier the compensation will usually need changing to avoid oscillation or gain peaking. In zero feedback amps the Hfe and Vbe become pretty critical and transistor matching is often required.... Some are rated as low noise, some for switching duty, some for RF uses and which may have a max voltage rating of only 12V. With fet's it's even worse as the range of Idss and pinch off voltage is huge even among devices of the same type and they often need selecting for Idss or to have a select-on-test resistor in the biasing...
If a circuit is pretty non critical in most ways, and has robust biasing, then a range of similarly rated transistors will work yes. eg in the line stages of something like a Quad 303 virtually any transistor that fits in the "small signal, >30V, >60MHz Ft, >50mA, >100Hfe" and of the correct polarity will work, but may well not be optimum. In many much more critical applications, changing a transistor that has been carefully chosen for its specific characteristics in that circuit will often result in the amp blowing up, and maybe taking out the speakers if it goes DC!
If a circuit is pretty non critical in most ways, and has robust biasing, then a range of similarly rated transistors will work yes. eg in the line stages of something like a Quad 303 virtually any transistor that fits in the "small signal, >30V, >60MHz Ft, >50mA, >100Hfe" and of the correct polarity will work, but may well not be optimum. In many much more critical applications, changing a transistor that has been carefully chosen for its specific characteristics in that circuit will often result in the amp blowing up, and maybe taking out the speakers if it goes DC!
We are talking I/U (current to voltage conversion) stage here not driver or output stages. There are no heavy loads at the R2R DAC current output! As a matter of fact the DAC itself would love to see a short circuit here.
We are dealing with a couple of mA currents at I/U conversion stage, which are very critical to the sound.
Nothing can really blow here as we are really dealing with tiny currents and small voltages <10V.
Even mixing up pins is not going to do any damage- happened to me a couple of times without any damage neither to the transistor nor to the I/U power supply.
I would not recommend to try any transistor that is a different breed other than bi-polar NPN, though e.g. PNP instead of NPN.
Arkless Electronics
17-09-2017, 18:44
We are talking I/U (current to voltage conversion) stage here not driver or output stages. There are no heavy loads at the R2R DAC current output! As a matter of fact the DAC itself would love to see a short circuit here.
We are dealing with a couple of mA currents at I/U conversion stage, which are very critical to the sound.
Nothing can really blow here as we are really dealing with tiny currents and small voltages <10V.
Even mixing up pins is not going to do any damage- happened to me a couple of times without any damage neither to the transistor nor to the I/U power supply.
I would not recommend to try any transistor that is a different breed other than bi-polar NPN, though e.g. PNP instead of NPN.
I was mainly wanting to point out to people that in general it is a very bad idea to attempt "transistor rolling" and at worst could release the magic smoke!! I'm sure you know what you are doing in the context of your own products;)
Chris just posted his impression on the DAC I Signature with the newly rolled I/U transistors. I installed 2 x 3 sockets and inserted his favorite sounding transistor which he determined by himself.
http://theartofsound.net/forum/showthread.php?52420-Production-Update
S
Greetings,
The review of SW1X DAC I STD Active I/V just got updated. One can read about the results of rolling different transistors in the I/U conversion stage. The updated review can be accessed here:
http://www.hifi-advice.com/blog/review/digital-reviews/spdif-dac-reviews/sw1x-dac-1/
S
Glad to see Slawa's ideas gaining traction. I believe he now offers this as a customer option on some of his DACs..
Like the appetite for tube rolling, this now gives enthusiasts yet another option to customise the sound to their liking. Differences easily match the scale of those achieved by tube rolling.
ovlov854
06-01-2018, 10:53
How did Germanium score?
As usual, they beat England on penalties! :lol:
Sorry, as you were.
Marco.
Considerably better than Francium ;)
:rfl:
Love it love it love it :laugh:
Powered by vBulletin® Version 4.2.3 Copyright © 2024 vBulletin Solutions, Inc. All rights reserved.