I’ve recently picked up a Border Patrol SE DAC. Cute because it’s technically NOS, non over sampling but it’s also uses a NOS new old stock Phillips TDA1543 chip from the 90s. These chips were the way to turn bits into music before Delta Sigma DACs hit the scene. They’re all R2R but it’s built into the Phillips chip so not like the R2R chips of today that have to build the ladders from scratch since these chips are no longer made.
You know when Schitt talk about they’re using of medical grade ladders? These were the chips designed for music. Used in super duper high end CD players of back in the day.
I wanted to talk about the NOS… the non over sampling part, not the NOS part. Within a few weeks I went from having most of my DACs be in the $200 dollar price range give or take. To having a couple of $1000 plus DACs, cause that’s how life works.
The other DAC I picked up was the Holo Cyan… not the Spring that everyone waxes their poetic shafts about. This is the Spings’s little brother but it’s still got the bloodline. I’ll write up reviews on both over the next few weeks but suffice to say that when it comes to DACs, listen first, measure second.
I spent about three hours Thursday just listening to music. Not critical listening, just enjoyment listening. They way it should be.
So yesterday I was listening to my Denafrips Pontus and to my ears the staging was really quite excellent but the separation was a little too exact and it made the presentation seem a little unnatural.
I looked at the DAC and factory setting had it in OS mode. Today I’ll go back to it now in NOS mode and I’m going to be pretty curious what I hear. It also had the filter set to SHARP which I changed to SLOW but I trust that will have a much more subtle effect on the sound.
I’ll also mention I’ve never looked at a DAC’s measurements and probably never will lol
I’m glad to hear that you are enjoying your DACs. I am extremely happy with mine and it’s one of those things that “I just can’t unhear.”
I have been a big proponent of the importance of DACs for a while and to invest in good electronics. I think mine plays a huge role in convincing me that the music is really there. The presentation is very natural and it excels in both soundstage and imaging.
I’m with you on the merits of the Border Patrol, I stumbled across it while researching the MHDT R2R DACS. I still want one or 2 of those also but the Border Patrol was interesting enough to me to add its name to my “purchase this immediately if you run across one used for a decent price list.” which is exactly what i did with no regrets currently.
I have NOT tried it with headphones yet, that’s really not why I purchased it, I am using it exclusively with speaker based systems. At some point I’ll try it w/ headphones and do a small review/opinion piece on it.
The interesting thing about the Border Patrol (Besides the R2R chip) is obviously the Tube rectifier power section. Actually the entirety of the power section is very meticulously implemented. But things like the chassis being made of copper (because we all know copper is less harsh than aluminum or folded/stamped steel) is that whatever voodoo they’re doing with it, it’s working on the analog sound that comes out of the RCA out.
Talk about the less is more approach. 1 input, 1 output. BTW the guy I bough it from had two spare tubes he threw in, they don’t seem to make a bit of difference in sound so I wouldn’t bother with tube rolling. They certainly make a difference when you engage the tube into the power circuit though.
To me the most interesting thing about the Border Patrol DAC is it’s lack of an output section it drives the line out directly from the DAC chip through capacitors.
The problem with doing that is it means the output impedance is ~2000 Ohms, so you probably want to couple it to an amplifier with an input impedance in the 50K+ Ohm range, that’s a lot of amps so it’s not an issue, but there are a few amps with input impedances down in the 20K Ohm range.
For the same reason you don’t want to use splitters on the signal, since it’s in effect putting the loads in parallel.
The TDA1543 DAC chip was commonly used in DIY designs a few years ago because it’s so easy to build a working DAC with one, you can still pick up variations of the DIY designs from Chinese manufacturers on EBay for very little. It’s also sometimes used in higher end DAC’s often with 16 or more in parallel.
I have it plugged into a Schitt Freya +. It’s sounding good so I’m assuming I’m okay. I also plugged it directly into a SB/BHC and again sounds delicious so assuming I’m okay there.
You point to the only issue that I’ve got with it so far. One set of outputs means switching cables back and forth whenever I want to listen to headphones or the system. Not a biggie but for a cable purist like me, splitting a signal is akin to sacrilege so I will have to live with swapping cables back and forth.
Anyway… not to leave my NOS Holo Cyan out of the picture. It also has the ability to switch OS to NOS. I know everyone says to go NOS but I like to listen to things for myself. And the reason I wanted the conversation is in hopes of finding out what is the NOS not doing to the signal that the OS does. In both of these DACs the NOS is just 'mo better. There’s more tangibility to the sound, I hate to just use non descriptive terms but it just feel more real. You know, like the first time you spank a lover and they tell you to do it harder and your brain short circuits for a second and during that second you whole perception of the universe changes.
The point of OS is to put the “noise” higher up into inaudible frequencies. But what if while doing that, we’re throwing the baby out with the bathwater? Music has noise, good noise, bad noise, good cholesterol, bad cholesterol. I don’t know enough to even formulate what I’m trying to say other than I am enjoying both of these DACs. In the case of the Border Patrol, I think it’s one of these peripheral pieces of hardware that’s just so different that it’s not only the NOS that’s creating the magic. In the case of the Holo Cyan, NOS is that same slap me harder kind of experience.
The original intent of oversampling was to increase the output resolution, by modulating the signal at a higher frequency than the sampling rate. The earliest Phillips chips were only 14 bits, so some of the first CD players used 4x Oversampling to add the missing 2 bits back in. This is a lot like what a DS DAC does, only with the majority of the bits directly decoded. I believe Schiit does this on the BF2 to increase the dynamic range.
That same technique was used by later DAC’s in CD players to smooth signals out, the distinction being there is no actual additional information in the signal dictating this, it’s basically applying a convolution of the up sampled signal.
The issue with doing that is you introduce time domain artifacts, if you were to look at an impulse response form an NOS DAC, the impulse is clean, from an oversampling DAC depending on the filter, you’ll see ultrasonic ringing either before or after (or both) around the actual impulse.
Lot’s of people seemed to think ringing before the impulse was “bad” in the late 90’s/early 2000’s, but that’s not really considered to be the case anymore.
Noise shaping is what’s used to shift noise into higher frequencies, again it’s technically just a convolution, though usually over a much larger set of samples, and it requires you have up sampled enough of a base sampling rate to actually shift the noise somewhere inaudible.
Signal processing is something I dabbled in decades ago, and I wouldn’t profess to be an expert, it’s a complicated field.
NOS DAC’s tend to have a relatively distinct character, there is a lot of debate on whether it’s a win or not and at what price point it becomes a win, there is a huge thread on NOS DAC’s over at SBAF if you want to wade through it.
Thanks for the overview. I guess it’s a field that you simply have to understand the science in order to explain it to the uninitiated because there are no bumper sticker level explanation. Like so many other things in audio, people like what they like and sometimes what you like puts you in the majority while other times it put you into the minority of the argument/opinion.
Then again sometimes it’s all about how well the stuff you buy synergizes together. This BP was the first DAC I’ve purchased knowing exactly what I was shooting for. Everything else before was purchased blind, trial and error and recommendations or reviews. There’s no substitute for first hand experience.
“Music noise” has nothing to do with it, the DAC doesn’t know which part of the input signal is considered by listeners to be “noise” or “music”, it just has to take everything that was digitized and turn it analog again. But the digitization process itself also introduces some noise when it converts a continuous line to a bunch of spaced-out samples. That noise has to be removed when you’re reversing the process at the DAC stage and turning the signal analog. If you don’t do that, you’re failing to perform a proper conversion from digital back to analog, and that’s what NOS DACs do: fail to convert digital to analog properly.
When it comes to the question of fidelity/accuracy, only objective, science and measurement-based arguments count. Enjoyability is another topic altogether, but fidelity or accuracy of reconstruction is a technical subject and has to be discussed with measurements and science. SBAF is not great at that, they often mix subjective and objective arguments and consider them both valid in the same discussion even though they don’t work for establishing the same kinds of conclusions. A better source for discussion of such a technical topic is obviously ASR:
The NOS DAC makes a sine wave look like a stair-step signal, which is a failure to reconstruct the original analog signal that was captured by the ADC. Once I saw this graph the NOS DAC case was closed for me. I don’t have the time or the money to experiment with whatever some people say is the “magic” or “special something” of NOS DACs, so I have to go with the hard engineering that shows me NOS = low-fi, bad reconstruction. It goes in the same bin with tube amps and vinyl for me, as lower-quality technology that should be left to the history books.
It’s 2020, a proper state-of-the-art DAC is a filtered delta-sigma DAC, and a proper output filter is a sharp roll-off linear-phase filter (F1 on Topping DACs that have 6 filter options, the “Standard” filter on the Micro iDSD or the “Measure” filter on the Nano iDSD BL).
How do you know what the original ADC was like? If we’re talking about mastering that happened in the 80s 90s what was the technology used by ADC then and how is throwing more data points at a sine wave a DAC can do now anything other than calculated guesses?
Based on this argument there should only be one ADC/DAC to assure the most accurate reproduction, clocking, everything else would be an attempt at recreating an unknown…
All the science in the world yet I’m enjoying the sound that I’m hearing from a stair stepped DAC. I’m not smart enough to say why I’m enjoying a NOS setting over a OS setting on a DAC. But I know I’m enjoying it more because it sounds right, more organic to me ATM based on what I’ve heard previously.
The stair stepping argument is specious.
The test shows one aspect of reproduction, the ability to reproduce a sin wave at a given frequency.
Smoothing the output in a DAC isn’t adding any more information, it’s like blurring an upsampled image, it looks smoother, but it isn’t higher resolution. It could sound better, or maybe just different, but that’s a different discussion.
You also don’t just listen to the output of the DAC, and everything else in the system (mostly the transducer) ends up smoothing that stair step anyway.
The primary argument for NOS DAC’s is they reproduce what was encoded.
Personally I have no Dog in this fight, there is too much discussion of what implementations sound like, whether it’s R2R vs Delta Sigma, or NOS vs OS, it should be about what a Signal chain sounds like.
Never ceases to amaze me that the objective vs subjective argument pops up yet again, considering the subject of this thread.
My question is simply this…if the person is enjoying what they are hearing, whether it measures well or not… why do people feel the need to bring up a polarizing topic that adds no value to the conversation?
IMO, shifting the topic of discussion from one’s enjoyment of their gear into a baited objective vs subjective argument is not only rude, it shows how little respect one has for other people’s enjoyment. There are plenty of different threads on this forum, as well as others, to have that discussion without shitting on one’s happiness.
While I get what the graphs try to show, they are horrible!
The 5kHz measurement is “broader picture” in a case where that does not give you ANY benefit (it is a sine wave, show 3 zero-crossings and done!)
After that (10 and 16kHz), it is just a mess that may aswell be down to aliasing in the display software.
This is a broad topic, judging by the title, nobody restricted it to subjectivist considerations. Furthermore, I didn’t say someone who enjoys something should stop enjoying it, I replied to a technically incorrect statement about noise removal (the DAC should remove the noise that the ADC introduced, nothing else; that is what its output filter is for) and then provided my own reasons for deciding (again subjectively, as are all decisions ultimately) to dismiss NOS designs as inferior. There is no point at which this was off-topic. Also there is no point at which any of this was like a punch in anyone’s face.
No. The responses shown side by side above have the same horizontal resolution. The defect of the NOS output is clearly visible when you compare them and is not caused by measurement resolution limitations.
You put it an analog signal of a well-known shape (like in the sine wave experiment) and measure what comes out after the ADC and DAC. This isn’t some mysterious dark magic. You can experiment with known signals and see what comes out. What comes out is an incorrect reconstruction that does not follow the Nyquist-Shannon sampling theorem. It’s not a matter of opinion, it’s maths.
Which is simply not relevant to the point the author tries to make there.
By keeping the horrizontal resolution fixed, he shows the frequency is changed (which is redundant as that is shown above the measurement series by the annotation that reads “x kHz”).
It is not. What is shown there may be down to the way the graphical output is rendered. As it is not an intensity graded image, it could mean anything.
What does this here tell you:
to many samples taken (1.66ms is too long of a range)
multiple triggers (the jitter in the signal generation causes the pulses not to line up perfectly, causing the spike pattern in the frequency spectrum)
Edit: It also shows a limitation of the algorythm used for the FFT (different topic)
With the measurement rig set to trigger externaly from the signal generator, only one sample is taken:
That is not applicable here. At all.
The WKS-Theorem simply states that to reconstruct a signal with the bandwidth f, you need 2f of sampling frequency.
See this is where it goes all wrong, because we’re not dealing with “well known shapes” we’re dealing with recorded and encoded music something that couldn’t get further from the unknowns. So how do we calculate for the unknown. I don’t disagree with the science, and I don’t disagree with mathematical formulas, what I am saying there’s a reason people seem to like NOS DAC output in certain implementations.
I couldn’t begin to explaikn the differences but I can see why less is more when the more is an interpolated calculation, maybe they don’t get it right regardless of how measurably better is is, does it sound better to a human ear?